Accident Metrics

[jlunseth]

We have had a few threads lately on the subject of takeoff and landing accidents, and takeoff, landing and approach techniques.  I got a notice of a WINGS program with links to the most recent Nall report, which is the 2015 report.  The full report breaks accidents down into categories, such as (1) takeoff and climb, (2) landing, (3) maneuvering, and (4) descent and approach.  There is an AOPA Scorecard report for 2016-17, but it has totals and not the full breakdown, so it is not as useful.  I recall that a few years ago there was an issue that accident cause determinations by the NTSB were not complete when the annual Nall report was coming out, so the Nall report, which used to be delayed by a year, is now delayed by two years.  

Looking at the '15 Nall report, the statistics are interesting and, I think, should factor in to how we fly our aircraft. (Unless specified, all the numbers are for GA fixed wing only, there are separate categories for commercial, and for rotorcraft in the report).

Total GA accidents of all types 967

Takeoff and climb accidents 108 total/ 19 fatal  11% of total accidents/17.5% fatality rate

Maneuvering 44 total/32 fatal     4.5% of total accidents/ 73% fatality rate

Descent/Approach 43 total/15 fatal     4.5% of total accidents/ 35% fatality rate

Landing 262 total/3 fatal     27% of total accidents/ 1% fatality rate

The percentage of total accidents should be looked at as just a relative comparison.  It depends as much on how many categories the accidents are divided into by the reporters as anything else, but when you see that, for example, takeoff and climb accounts for 11% of accidents, and descent and approach is 4.5%, you get a relative measure of which flight regime causes the most problems. Landings is the winner at 27%, the highest of all the categories, but also far and away the least likely to be fatal.

The categories are broken down further, and that is the interesting part.  Takeoff and climb have 8 subcategories, but of those, loss of control accounts for 40 accidents and stall or settle on takeoff accounts for 36.  The others are relatively insignificant.  Further, loss of control on takeoff resulted in 4 fatalities, so 10% of the time that type of accident is fatal.  Stalls resulted in 9 fatalities for a 28% fatality rate.  Stalls on takeoff are the biggest danger.  LOC on landing results in alot of accidents, but a small fatality rate, somewhat similar to the LOC rate on takeoff although the takeoff fatality rate is higher. 

Maneuvering results in a moderate accident rate, but a very high fatality rate - stalls again being the major risk, causing 55% of maneuvering accidents with a very high 91% fatality rate.  Descent and approach, also a moderate accident rate and what I consider to be a high fatality rate at 35%.

The subcategories for descent and approach are what is interesting.  There are four subcategories, (1) stalls/spins, (2) collisions, (3) deficient instrument approach procedures, and (4) loss of power.  For purposes of deciding what is the safest way to conduct a standard approach and landing, you can kick out deficient instrument approach procedures and focus on the other three.  The leading cause is stalls/spins, which accounted for 13 accidents, 5 of them fatal, for a fatality rate of 38%, which is uncomfortably high.  Collisions were second accounting for 11 accidents, 2 fatal.  And loss of power?  Guess what.  9 accidents and zero fatalities.  

So much for the ole "tight pattern" propaganda.  Tight patterns are advocated so that, in the event of an engine failure in the pattern you can "make it to the runway."  But they are rarely if ever fatality.  What causes pilots and their passengers to die at an uncomfortably high rate, is a stall/spin during the approach.  Your odds of that happening at all are slightly higher than a loss of power, and your odds of dying from it are 35 times higher than a loss of power.  

The overall accident rate is down, but pilots are still the biggest cause at 73.5%.  Here are links to the 2015 Report and the 2016-17 Scorecard.  http://bit.ly/2JL1nxM  http://bit.ly/1617GAscore 

In general, when you look at the accident rate for stalls during maneuvering, during takeoff and climb, and during descent and approach, the biggest risk of dying is from stalls.  Fly your takeoffs, climbs, approach, and descent to avoid stalls and the fatality risk goes way down.

[FloridaMan]

Don't forget that incidents that occur during different phases of flight carry a different likelihood of resulting in what gets reported as an accident. For instance, if you lose control during maneuvering and go into a spin, the odds are that you'll recover in most circumstances where you have altitude to spare. 

https://medium.com/@penguinpress/an-excerpt-from-how-not-to-be-wrong-by-jordan-ellenberg-664e708cfc3d 

[Yetti]

39 minutes ago, jlunseth said:

So much for the ole "tight pattern" propaganda.  Tight patterns are advocated so that, in the event of an engine failure in the pattern you can "make it to the runway."  But they are rarely if ever fatality.  What causes pilots and their passengers to die at an uncomfortably high rate, is a stall/spin during the approach.  Your odds of that happening at all are slightly higher than a loss of power, and your odds of dying from it are 35 times higher than a loss of power.  
 

Whoa there.  Lots of stats and facts and then a jump to tight patterns cause stall spins.  Probably need some data to make that leap.

[jlunseth]

18 minutes ago, Yetti said:

Whoa there.  Lots of stats and facts and then a jump to tight patterns cause stall spins.  Probably need some data to make that leap.

Why whoa?  There is a lot of data there.  Loss of power on approach happened 9 times and caused no, as in zero, fatalities last year.  So where is the excuse to fly a tight pattern “so you can make it to the airport if the engine quits?”  What causes the stall/spins is tight patterns, causing tight banked turns downwind to base and base to final.  

One could say that the numbers are small, so how statistically signficant is 9?  The issue with that is that the Nall report has come out since the 50’s, and while the total accident rate has come down, the numbers from prior years say the same thing.

The “ringer” as far as I am concerned, in the Nall reports, is that they include experimental in the category of GA.  They used to do that, but break out the numbers for experimental and the numbers were very high compared to the rest of GA.  

[jaylw314]

1 hour ago, jlunseth said:

In general, when you look at the accident rate for stalls during maneuvering, during takeoff and climb, and during descent and approach, the biggest risk of dying is from stalls.  Fly your takeoffs, climbs, approach, and descent to avoid stalls and the fatality risk goes way down.

While I don't necessarily disagree with your conclusion, I do take (professional, not personal) exception to your approach.  It's important to remember that as you break down generic grouped data into specific broken-down data, you're reducing the power of your conclusions because of the (by definition) reduction in group or sample sizes.  In addition, the broken-down data becomes more vulnerable to observer bias and fallacious conclusions. Maybe a better way of saying that is that the more useful the conclusion you want to make, the less likely your conclusion is to be correct (and vice versa).

While the overall grouped data and the first couple layers might be good enough to base a real-life decision on, the most broken down data probably should not be used to base real-life decisions on in isolation

[Yetti]

41 minutes ago, jlunseth said:

Why whoa?  There is a lot of data there.  Loss of power on approach happened 9 times and caused no, as in zero, fatalities last year.  So where is the excuse to fly a tight pattern “so you can make it to the airport if the engine quits?”  What causes the stall/spins is tight patterns, causing tight banked turns downwind to base and base to final.  

One could say that the numbers are small, so how statistically signficant is 9?  The issue with that is that the Nall report has come out since the 50’s, and while the total accident rate has come down, the numbers from prior years say the same thing.

 

That is the breakdown of your jump.   What if you are flying a continuous turning approach because it is less likely to stall spin than a box pattern?   Define your definition of "tight banked turns"   Could we agree that standard rate turn at pattern speeds or less than 25 degrees is not a "tight banked turn"?

Certified or  experimental is not part of this discussion

Edited October 6, 2018 by Yetti

[MIm20c]

I agree with @Yetti, it doesn’t matter how large the pattern is flown if the base to final turn is late or sloppy that can lead to problems. 

Otherwise thanks for the link and info. I really like digesting the data. 

[TNIndy]

Thanks for the information

[201er]

“So much for the ole "tight pattern" propaganda.  Tight patterns are advocated so that, in the event of an engine failure in the pattern you can "make it to the runway."  But they are rarely if ever fatality.  What causes pilots and their passengers to die at an uncomfortably high rate, is a stall/spin during the approach.  Your odds of that happening at all are slightly higher than a loss of power, and your odds of dying from it are 35 times higher than a loss of power.“

 

I think it’s the opposite solution. If pilots flew tight patterns all the time, they’d be more proficient at making turns without stalling. Seems like a lot of stall accidents happen when someone is put outside their element. ATC tells them to keep it tight or the first time they took all their fat friends flying. If you are used to safely performing tight patterns as a matter of habit then what can be outside your element? Making a wide pattern? Hardly a calamity that’s going to send the pilot for a loop.

If pilots would learn to control angle of attack rather than thinking in terms of speed we’d go a long way. Every stall accident is the result of improper angle of attack control. 

Speed is too confusing and too indirect of a measure of angle of attack. There are so many different speeds happening simultaneously. Indicated speed, true speed, ground speed, vertical speed... when pilots get in the habit of seeing them as equal or close enough, they can make a big mistake when they turn out not to be! Tailwind, altitude, density altitude can all spread them apart and even more so when combined.

Most pilots don’t practice steep turns, precise angle of attack control, and high angle of attack flight sufficiently so when it happens they are less prepared. If we believe that currency, proficiency, instrument currency, etc come from practice and frequency, then so does stall prevention. Flying tight patterns correctly is a chance to practice and be more prepared for when that type of flying happens unexpectedly.

Just my four half pennies.

Edited October 6, 2018 by 201er

[Shadrach]

3 hours ago, jlunseth said:

Why whoa?  There is a lot of data there.  Loss of power on approach happened 9 times and caused no, as in zero, fatalities last year.  So where is the excuse to fly a tight pattern “so you can make it to the airport if the engine quits?”  What causes the stall/spins is tight patterns, causing tight banked turns downwind to base and base to final.  

One could say that the numbers are small, so how statistically signficant is 9?  The issue with that is that the Nall report has come out since the 50’s, and while the total accident rate has come down, the numbers from prior years say the same thing.

The “ringer” as far as I am concerned, in the Nall reports, is that they include experimental in the category of GA.  They used to do that, but break out the numbers for experimental and the numbers were very high compared to the rest of GA.  

Pull the type specific data. How many Mooney’s are in pattern stall/spin accidents?

[Shadrach]

41 minutes ago, 201er said:

“So much for the ole "tight pattern" propaganda.  Tight patterns are advocated so that, in the event of an engine failure in the pattern you can "make it to the runway."  But they are rarely if ever fatality.  What causes pilots and their passengers to die at an uncomfortably high rate, is a stall/spin during the approach.  Your odds of that happening at all are slightly higher than a loss of power, and your odds of dying from it are 35 times higher than a loss of power.“

 

I think it’s the opposite solution. If pilots flew tight patterns all the time, they’d be more proficient at making turns without stalling. Seems like a lot of stall accidents happen when someone is put outside their element. ATC tells them to keep it tight or the first time they took all their fat friends flying. If you are used to safely performing tight patterns as a matter of habit then what can be outside your element? Making a wide pattern? Hardly a calamity that’s going to send the pilot for a loop.

If pilots would learn to control angle of attack rather than thinking in terms of speed we’d go a long way. Every stall accident is the result of improper angle of attack control. 

Speed is too confusing and too indirect of a measure of angle of attack. There are so many different speeds happening simultaneously. Indicated speed, true speed, ground speed, vertical speed... when pilots get in the habit of seeing them as equal or close enough, they can make a big mistake when they turn out not to be! Tailwind, altitude, density altitude can all spread them apart and even more so when combined.

Most pilots don’t practice steep turns, precise angle of attack control, and high angle of attack flight sufficiently so when it happens they are less prepared. If we believe that currency, proficiency, instrument currency, etc come from practice and frequency, then so does stall prevention. Flying tight patterns correctly is a chance to practice and be more prepared for when that type of flying happens unexpectedly.

Just my four half pennies.

Agree! The guy that flies 2.5 mile pattern legs on a regular basis will be at greater risk in situations that call for tighter patterns and or maneuvering.

Edited October 6, 2018 by Shadrach

[Yetti]

8 minutes ago, Shadrach said:

Pull the type specific data. How many Mooney’s are in pattern stall/spin accidents?

3 or 4 Mooney I can think of this year.

[carusoam]

You would think the Nall report could be Mooney specific at the push of a button. Is this still the 90s?  Or do they not want us to use and discuss the data...  

Somewhere around here, somebody posted a thread that sifted all the accident reports.  You could easily see what your Mooney cadre were doing...

Thinking about the box pattern...

1) In and out of banks allows the pilot to do one step at a time... turn, then configure, then trim ... repeat...  (old style of training)

2) in and out of banks twice, doubles the opportunity to over bank... while going slowly...

3) a continuous turn makes it a bit easier to roll out on final... no overshooting /correction required... similar to driving through an intersection while turning 90°.  

 

4) set the bank angle, watch the rate of turn, judge wether the turn will be completed in time or not, continuously... take out the bank smoothly....

5) the cost... multi tasking is required... turn, descend, configure, trim... all at the same time...

6) don’t forget, one configuration change... gear down, full flaps, trim before starting the turn...  (whatever full is for your needs)

7) I use gear down and first notch before being in the pattern... (Long body)

8) Stall speed improves with more flaps...

9) Removing steps simplifies the procedure allows the pilot to keep the Angle of bank and the AOA better controlled...

10) Nothing like being late to start the turn to final... except for being fast at the same time... the brain wants to raise the nose to slow down... and bank steeper to make the turn... then use the rudder to align the nose... the result... predictably being slow, steep banked, and cross controlled... before the retracing wing falls out...

Engine failures have mostly been climbing to tpa, around here... very stressful on the engine... compared to during landing... in most cases landing straight ahead has been successful... even with trees...

moral of the story... keep it simple to avoid bad things...  wax the wing leading edges to go faster... try to avoid hard objects... go around early.... if you fail to go around early... scrap the plane... and walk away...

PP thoughts only, not a CFI...

Best regards,

-a-

 

[INA201]

     I agree with the assertion that we as pilots are the largest cause of accidents.  I would also add that deferred maintenance and or having a thorough knowledge of all parameters of maintaining our aircraft in excellent flying shape should add to the pilot error statistic.  The maintenance side of the pilot error statistic can include IFR without a backup attitude indicator all the way to not replacing hoses in a timely fashion, or even a proper engine monitor and treatment of the engine.  There is virtually no way to statistically determine the effect that maintenance has on mechanical malfunction.  We can assume that maintenance is a "known" variable.  I'm making the assumption that some 85% of all accidents are pilot caused based on maintenance being an added factor.  This 85% is just a thrown out number because it certainly could be less or more.

     I fly patterns differently at different airports depending on the situation.  You may be given a four mile base leg by ATC, straight in, downwind leg, or fly a more normal pattern at an uncontrolled field.  One thing I do is to be set up for landing as soon as practicable.  When making turns at low power and speed I will coordinate my turns with a descent.  If pattern altitude is 1000 AGL I'll turn downwind to base while descending to 600+-.  I will then make my final turn while descending to match the VASI, PAPI etc. all while maintaining airspeed.  I'm not sure if this is appropriate but I've just always done it this way as a habit.  In my opinion a level turn while going slower with flaps and gear out is a scary maneuver.  Does anyone else do this or feel this way? 

      

[jaylw314]

2 hours ago, INA201 said:

     I agree with the assertion that we as pilots are the largest cause of accidents.  I would also add that deferred maintenance and or having a thorough knowledge of all parameters of maintaining our aircraft in excellent flying shape should add to the pilot error statistic.  The maintenance side of the pilot error statistic can include IFR without a backup attitude indicator all the way to not replacing hoses in a timely fashion, or even a proper engine monitor and treatment of the engine.  There is virtually no way to statistically determine the effect that maintenance has on mechanical malfunction.  We can assume that maintenance is a "known" variable.  I'm making the assumption that some 85% of all accidents are pilot caused based on maintenance being an added factor.  This 85% is just a thrown out number because it certainly could be less or more.

     I fly patterns differently at different airports depending on the situation.  You may be given a four mile base leg by ATC, straight in, downwind leg, or fly a more normal pattern at an uncontrolled field.  One thing I do is to be set up for landing as soon as practicable.  When making turns at low power and speed I will coordinate my turns with a descent.  If pattern altitude is 1000 AGL I'll turn downwind to base while descending to 600+-.  I will then make my final turn while descending to match the VASI, PAPI etc. all while maintaining airspeed.  I'm not sure if this is appropriate but I've just always done it this way as a habit.  In my opinion a level turn while going slower with flaps and gear out is a scary maneuver.  Does anyone else do this or feel this way? 

      

I don't think I agree that deferred maintenance is a large contributing factor.  I can't think of many equipment failures that would always result in an accident--even the loss of vacuum in IMC should be a recoverable failure if we keep current with partial panel skills.  Complete and total catastrophic failures are far more rare, and then there's the question of whether deferred maintenance was actually the cause, since the phrase "deferred maintenance" has no consistent definition--not doing a 500-hour IRAN on a magneto would probably be obvious, but what about the whole "running a motor past TBO" thing?  Reading accident reports, there are plenty of shocking cases of planes that shouldn't be flying, yet the accident was associated with a failure that should have been recoverable by the pilot.

If a pattern of inappropriately deferred maintenance is statistically associated with pilots who have accidents, I suspect it is simply that--an indicator of a pilot who is more likely to make poor decisions and errors.  Unfortunately, I can't think of any easy way to measure such a thing.

[Hank]

5 hours ago, INA201 said:

      I fly patterns differently at different airports depending on the situation.  You may be given a four mile base leg by ATC, straight in, downwind leg, or fly a more normal pattern at an uncontrolled field.  One thing I do is to be set up for landing as soon as practicable.  When making turns at low power and speed I will coordinate my turns with a descent.  If pattern altitude is 1000 AGL I'll turn downwind to base while descending to 600+-.  I will then make my final turn while descending to match the VASI, PAPI etc. all while maintaining airspeed.  I'm not sure if this is appropriate but I've just always done it this way as a habit.  In my opinion a level turn while going slower with flaps and gear out is a scary maneuver.  Does anyone else do this or feel this way? 

I generally turn base when the runway is 45° over my shoulder, and with 1000agl pattern thats generally around 800 agl. Final is usually right around 500 agl. Once I drop the gear abeam my intended point of landing and reduce throttle, I am descending and never level off; my turns are often standard rate but always 30° bank or less. Speed is 90mph downwind and base, rolling wings level on final at 85, crossing the numbers at 75 - 5 mph for every 300 lb below gross I am at that time. Minimizes float and allows for a gentle touchdown when I don't level off a couple of feet too high . . .

[INA201]

4 hours ago, jaylw314 said:

I don't think I agree that deferred maintenance is a large contributing factor.  I can't think of many equipment failures that would always result in an accident--even the loss of vacuum in IMC should be a recoverable failure if we keep current with partial panel skills.  Complete and total catastrophic failures are far more rare, and then there's the question of whether deferred maintenance was actually the cause, since the phrase "deferred maintenance" has no consistent definition--not doing a 500-hour IRAN on a magneto would probably be obvious, but what about the whole "running a motor past TBO" thing?  Reading accident reports, there are plenty of shocking cases of planes that shouldn't be flying, yet the accident was associated with a failure that should have been recoverable by the pilot.

If a pattern of inappropriately deferred maintenance is statistically associated with pilots who have accidents, I suspect it is simply that--an indicator of a pilot who is more likely to make poor decisions and errors.  Unfortunately, I can't think of any easy way to measure such a thing.

You may be right. There is no way to statiscally make a judgement here. 

I was in a partnership at one time where the guys wanted to get an annual from the “pencil whip guy.” It was a $600 annual.  He had rebuilt the wrong mag and the old one died while departing with the good mag running rough. There was no accident but it was definitely our fault for going to this guy.  Fortunately there was no statistical accident.  Anyway, there are some risks that are not able to be written or proven so I defer.

[DustinNwind]

I noticed a lot of people associating a tight pattern with high bank angle. Remember slower speed, less bank can have a better result than high speed and high bank angle. The faster you go the wider your turn is going to be, thus just slow down, do a standard rate turn, land her, and make the tower and traffic happy. 

[Hank]

2 hours ago, DustinNwind said:

I noticed a lot of people associating a tight pattern with high bank angle. Remember slower speed, less bank can have a better result than high speed and high bank angle. The faster you go the wider your turn is going to be, thus just slow down, do a standard rate turn, land her, and make the tower and traffic happy. 

That's what I do. In the Cessna, I flew pretty close to ½ mile out, over town; when I bought the Mooney, I noticed I was flying about 3 blocks further out. I rarely bank more than standard rate.

[Stephen]

1 hour ago, DustinNwind said:

I noticed a lot of people associating a tight pattern with high bank angle. Remember slower speed, less bank can have a better result than high speed and high bank angle. The faster you go the wider your turn is going to be, thus just slow down, do a standard rate turn, land her, and make the tower and traffic happy. 

^ Great advice if you have been driving a 172 around the pattern at a much higher airspeed than norms for the platform.

But for Mooney's...

Mooney's are  an *on-speed* aircraft and will clearly let you and insurer know when excessively fast on final due to excessive float (overruns, porpoising & prop-strike due to trying to force the aircraft nose-low onto the runway) ...or  your insurer/next of kin know when excessively slow. Systematically speaking, pilots who are flying Mid/Short body M20s are doing it at 100/90/80 MPH ("ish" +/- for weight). If that is incompatible with other aircraft that have a 20MPH lower stall speed and thus tighter patterns... there is one choice to avoid crowded pattern contention  and that is the same thing that jets do... fly bigger pattern which allows both for airframe appropriate speeds and more gentle turns. 

Making traffic and tower happy isn't killing Mooney occupants.. things like slow airspeed combined with higher rate base-to-final turns & departure stalls are definitely killing distracted pilots. So, you don't have new Mooney pilots staying off-speed in either direction for long periods as training and the laminar flow wing and slippery aerodynamics will respectively correct.... or tend to hand your wallet (possibly with your (_!_) attached) as corrective feedback.  The only time that I can think of Mooney's en-masse going faster than the above mentioned speeds  "in the pattern" is when they are not in the pattern and rather on an instrument approach and trying to keep 90Kts + on long-medium final for approach traffic flow compatibility. 

[DustinNwind]

8 hours ago, Stephen said:

^ Great advice if you have been driving a 172 around the pattern at a much higher airspeed than norms for the platform.

But for Mooney's...

Mooney's are  an *on-speed* aircraft and will clearly let you and insurer know when excessively fast on final due to excessive float (overruns, porpoising & prop-strike due to trying to force the aircraft nose-low onto the runway) ...or  your insurer/next of kin know when excessively slow. Systematically speaking, pilots who are flying Mid/Short body M20s are doing it at 100/90/80 MPH ("ish" +/- for weight). If that is incompatible with other aircraft that have a 20MPH lower stall speed and thus tighter patterns... there is one choice to avoid crowded pattern contention  and that is the same thing that jets do... fly bigger pattern which allows both for airframe appropriate speeds and more gentle turns. 

Making traffic and tower happy isn't killing Mooney occupants.. things like slow airspeed combined with higher rate base-to-final turns & departure stalls are definitely killing distracted pilots. So, you don't have new Mooney pilots staying off-speed in either direction for long periods as training and the laminar flow wing and slippery aerodynamics will respectively correct.... or tend to hand your wallet (possibly with your (_!_) attached) as corrective feedback.  The only time that I can think of Mooney's en-masse going faster than the above mentioned speeds  "in the pattern" is when they are not in the pattern and rather on an instrument approach and trying to keep 90Kts + on long-medium final for approach traffic flow compatibility. 

I agree " slow airspeed combined with higher rate base-to-final turns & departure stalls are definitely killing distracted pilots" . I'm not instructor but I do know that 80mph and a standard rate turn will give you tighter turns than 100/90mph turns. 

[Stephen]

Problem is for normal, non-short field approaches, 80mph is your short-final speed, not a normal pattern base leg speed so you are turning at 90 not 80. Should have clarified the 100/90/80 are relatively my downwind/base/final speeds and over the numbers at 75. I find it works well.

[steingar]

No big surprises.  Loosing it on takeoff is often fatal, since you're close to the ground with lots of energy.  Loosing it on landing is less often fatal.  Getting into a stall/spin is always a bad idea.

Yeah, we should all proficient in this and that.  But we all aren't.  Few of us fly as often as we'd like, and without that level of currency proficiency suffers.  Says me I'll do what is most likely to keep me out of bad accidents, like flying a wider pattern.  If ATC wants me to keep it tight I'll keep my speed up.  If ATC wants me to keep it tight and it's a dinky little airstrip I'll go elsewhere.

 

And no matter what I"ll keep the damn ball centered.  If more pilots did that there'd be way fewer accidents.

[Yetti]

U turn to land with decreasing energy.  downwind threshold at 120mph.  

 

[Stephen]

Purdy.